Baseball player and philosopher Yogi Berra said “if you don’t know where you are going, you’ll end up someplace else.” Adopting an AI product for a boutique law firm feels like this. The New York Times is calling for disruption to the practice of law. There are major product launches almost weekly. It’s hard to know how the practice of law will change and what your clients will expect from you in the future.
With such uncertainty and change, the first step must be to slow down.
To ask yourself a question: How can I validate my AI strategy before spending my budget?
Computer science offers a useful guiding principle in the classic hill climbing problem. You are trying to reach the top of the highest hill, but you can only verify your current height. You cannot see the full landscape around you. The simple approach is to start from a random point and test if the adjacent space is higher. If so, you continue moving up that path to the top. The risk is that you end up stuck on a local peak, well short of the highest top. So computer scientists have come up with techniques for a better approach. One of these techniques is to sample random areas, collect information to identify which is the highest hill, then begin the ascent.

To define an AI strategy today, you need to think about not only which hill you’re climbing, but also who will help you climb it. A sherpa.
The good news it that there are dozens of them available. This month alone saw major announcements for the Microsoft Word Legal Agent, Claude for Legal, and a slew of product releases from Harvey, Docusign, and LexisNexis. The problem is that choosing the wrong sherpa means you will go up the wrong hill, or won’t make it to the top. Switching to a new hill will be costly.
The way to optimize the decision is by sampling.
A key to sampling is to delay signing long-term contracts.
The major Legal AI vendors offer 1-2 week pilot periods. The foundation model vendors (Claude Teams or ChatGPT Business) offer flexible monthly subscription terms. In a few months, you can sample vendors so that you have confidence in your goals, preferences, and who can help you. Only then commit your budget and start the climb.